MEMS Switches Meet the Challenges of 5G Infrastructure
by Menlo Micro
RF switches may not get the same level of attention as their more technologically glamourous counterparts like RFICs and MMICs, mmWave SoCs, and phased-array antennas, but they arguably play an equally important role in wireless infrastructure. That’s especially true today, as wireless carriers add new bands to the large number already in service. And as just so much room is available in these subsystems, it’s essential for every component to meet increasingly stringent demands. So, it should be no surprise that research and development of advanced materials, processes, and other efforts continue apace, with the most formidable being MEMS technology.
Today’s base stations have bulky, unreliable and power-hungry electromechanical relays (EMRs) that drive up power consumption, generate heat, take a considerable amount of space, and typically require coaxial interconnects. In some cases, this requires active thermal management that is necessary to ensure that they will survive. These components not only significantly increase system cost and complexity, but also aren’t well suited to small platforms such as small cells for which issues such as power consumption, size, and weight are critical metrics. And as systems shrink thanks to the “digitalization” of previously analog functions, EMRs have become a significant obstacle in the path of realizing 5G.
What’s long been needed is an alternative that does not suffer from the traditional electromechanical switch’s limitations while also providing better overall performance in a fraction of their size, with negligible power consumption and much longer operating lives. MEMS technology satisfies these requirements and many others, as well.
The closest competitor to MEMS are solid-state switches that are small, fast, and reliable but are also comparatively power inefficient and, to a lesser extent than their traditional electromechanical counterparts, generate heat that can require heat sinks and complex thermal management. Semiconductors are never fully “off,” and the resulting leakage currents waste power. Researchers throughout the world have been attempting for years to overcome the shortcomings of both traditional electromechanical and solid-state RF switches, but the result, until recently, has been a series of compromises rather than an ideal solution.
New (and Challenging) Applications for MEMS
MEMS technology has been employed for decades for a wide variety of applications, but for RF switching it is comparatively new, and a history of its development over two decades is littered with more than a dozen companies that tried and failed to solve the vexing challenges this technology presents.
MEMS switches have movable parts (a fundamental feature of MEMS in general) like their EMR counterparts, but in every other way a MEMS switch is almost entirely different. They are microscopic, are created on a wafer in a semiconductor fab, and have virtually no mass (Figure 1). In fact, they’re also different from any other MEMS application, so much different that developing and commercializing a reliable MEMS switch product has eluded researchers for more than 40 years. These efforts consumed the resources of many companies, causing nearly all of them to exit the field before they could achieve commercial success.
The extraordinary efforts toward making MEMS RF switches a reality failed for several reasons, such as limited resources, a lack of fabrication facilities and control of the fabrication process. Most important, they were hindered by the inability to find materials to solve the unique challenges required to fabricate MEMS switches with the reliability, manufacturability, performance, and low cost required of a commercial product. In short, even though a MEMS switch could result in a paradigm change for switching, there is just so much money that an organization can expend without success before giving up entirely.
How They Work
An ohmic MEMS switch operates in the following way: When a DC voltage applied to the gate electrode of the switch reaches a specific value, the beam (or cantilever) bends downward due to electrostatic force and meets the contact below it—connecting the input to the output. (Figure 2a). When the DC voltage falls below this value, the beam returns to its original position (Figure 2b).
All this occurs in microseconds. The key challenge is how to manufacture these tiny devices at scale and configure them to withstand thousands of volts, tens of amps, and kilowatts of RF power —and do so for years, or even decades, without failure. Packaging is another essential contributor to making a MEMS switch reliable because it must maintain a stable environment for the switch to operate in, while also being able to scale for low-cost manufacturing.
In addition, while all switches employ a metal beam that actuates, MEMS switch actuators are so small their mass is negligible. This, in part, means they can deliver constant performance even during extreme shock and vibration, where most electromechanical relays subjected to the same set of extreme vibration and shock testing would fail.
For example, multiple electromechanical relays can be replaced by MEMS switches housed in a 2.5 x 2.5 x 0.9 mm chip-scale package and even when employed in huge switch matrices, consume less DC power than a single electromechanical switch. A MEMS switch not only switches 1000 times faster but is also at least 90% smaller, consumes almost no power, and can survive more than three billion switching operations even when handling relatively high RF power levels.
From GE to Menlo Micro
The solution that made MEMS switches viable for RF applications is the result of intensive research conducted by General Electric. The impetus for MEMS switch development at GE in 2004 was the company’s desire to find an alternative to traditional mechanical relays for remote programmable, very-high-power circuit breakers. The ideal solution would handle high power at the speed of solid-state technology and have the ability to perform reliably for decades of life, but without the losses associated with solid-state devices.
Existing ohmic MEMS switches from multiple vendors were considered, but at the time, a lack of reliability under harsh environmental conditions ruled them all out. No other technologies checked all the boxes either, so GE decided to embark on an effort to create their own ohmic MEMS switches from scratch.
Ohmic MEMS switches have two primary failure mechanisms: metal fatigue and contact wear. GE researchers determined that while metals are excellent conductors, they are not good spring materials for the cantilever because they deform over time, especially with variations in temperature. Thus, GE began an exhaustive process of materials evaluation that led to a proprietary fabrication process and a proprietary electrodeposited alloy.
The result was a mechanical actuator that combined mechanical properties near those of silicon with the conductivity of a metal. Metal alloys that ensure an absence of metal fatigue that has effectively eliminated this fundamental problem as a failure mechanism. It can deliver stable performance after more than 3 billion on/off operations, a figure that will soon likely rise to more than 20 billion operations.
GE’s success led the company to create a new company in 2016 that is now Menlo Micro in the hope that it would expand on its efforts to create viable RF switches. The advanced alloys GE created are the key components used by Menlo Micro to fabricate switches that can handle kilowatts of power (and therefore high temperature operation) over decades of useful life.
Menlo Micro has created a growing product line of MEMS switches called the Ideal Switch. Working with Corning, Menlo Micro demonstrated the integration of an innovative Through Glass Via (TGV) packaging technology for universal MEMS switches, which allows the Ideal Switch to be housed in tiny wafer-scale packages that reduce package parasitics by more than 75%. This allows the current switch portfolio to operate from DC to 50 GHz, with upcoming designs pushing past 60 GHz. Menlo Micro has reduced the size of its products by more than 60% compared with wire-bond packaging, which increases channel density while reducing size, weight, power, and cost.
Another benefit realized by this approach is the ability of MEMS switches to offer exceptional thermal performance in harsh operating environments with minimal variation in RF performance over temperatures from -40° C to +150° C. Menlo’s MM5140 (Figure 3) has operational performance data over this temperature range while achieving an insertion delta variation of 0.05 dB. Performance characteristics are shown in Table 1. Menlo’s Ideal Switches have also been used in extremely cold applications ranging from liquid nitrogen baths at -196° C to quantum computing dilution fridges at temperatures as low as 10K.
The extremely low mass of these MEMS switch components also results in reliability levels far exceeding those of standard electromechanical switches, enabling superior environmental performance and resistance to shock and vibration. A Menlo Micro switch, for example, exceeds the IEC 60601/60068 standard and passes MIL-STD 810G/H stresses for vibration and shock.
With the proliferation of frequency bands, having flexible radio architectures that can switch to multiple bands “on demand” becomes critical. Menlo Micro RF switches are well suited for these applications because they have ultra-low losses and introduce near zero distortion (non-linearity) to the circuit. They also provide orders of magnitude lower distortion than solid-state switches, which helps reduce the thermal footprint of the system and decreases the number of components and the overall volume by more than 90% (Figure 4).
In addition, many test and measurement systems need to switch hundreds or even thousands of RF signal paths when testing different products with multiple instruments. This can become an extremely complicated problem because the density of RF switching and the performance required creates massive RF switch matrices that require racks of equipment. Menlo Micro RF switches are well suited for these use cases because they combine high RF performance with low power consumption, minimal distortion, and a much smaller footprint.
When compared to an electromechanical relay used in such applications, the volume occupied by a single Menlo Micro switch would be just 25 mm3 while a relay would consume up to 30,000 mm3. When arrays of switches are required to accommodate 128 x 128 or 256 x 256 signals, the size of the system will grow very quickly. Operating lifetime is more than 3 billion operations versus 10 million operations for the typical RF relay and switching speed is less than 5 µs versus 5 ms.
Not Just for RF
While other MEMS switch technologies have been designed to switch RF signals, only the Ideal Switch can also manage AC and DC power. This opens the potential markets and applications much broader than previous MEMS switch efforts. Some of the applications originally envisioned by GE engineers will also soon be possible with switches being manufactured by Menlo Micro. The number of power relays, either solid-state or mechanical, being sold worldwide is in the billions. These power relays are unique as they need to handle very large loads of kilowatts and higher and operate reliably for very long periods. Applications can range from electronic control relays to AC/DC contactors and miniature circuit breakers, all of which are prevalent in industrial automation, battery management, vehicles, home automation, and many others.
For example, Menlo Micro’s Ideal Switch technology has already proven to be scalable with a 200V/20A MEMS power relay and testing shows that even higher levels can be achieved. Some benefits exhibited in other market segments are also available in the power sector as well. For example, low losses can eliminate the need for bulky heatsinks, reducing size and volume by up to 90%.
Fast switching speed and high reliability will make them stand out when compared to traditional mechanical relays and contactors. With the increased trends towards Industrial IoT and the “electrification of everything,” the need for miniaturized, low-power, and reliable power relay technology will continue to grow.
Telecommunications systems, medical equipment, and various test and measurement systems require switching of dozens or even hundreds of signal paths. These can be either DC or low-frequency AC signals with high voltages up to 1000 V and a few amps of current. Electromechanical or reed relays have been the switch of choice for decades because high-density solid-state solutions cannot handle the power while meeting low-leakage and high-isolation requirements. However, Menlo Micro switches not only meet these requirements but do so in a fraction of the footprint with minimal power consumption.
The history of MEMS RF switch development is a long one, and after enormous amounts of time and effort, many researchers concluded that even though MEMS technology had enormous promise, it also posed enormous challenges. Not surprisingly, many researchers who had spent much of their careers trying to overcome them simply gave up. The problem was not only making this technology work for RF switching, but also doing so in a way that it could operate in hostile environmental conditions.
Fortunately, the efforts of General Electric and today Menlo Micro show that MEMS RF switches can not only be achieved, but fabricated in a similar fashion to standard silicon CMOS that allows them to be manufactured in high volumes and scaled in voltage, current, and power handling. These are the early days in the evolution of this technology, and the company’s research shows that much more can be achieved in the coming years.